Hi everyone, I have a model which blows up for certain initial points when using jitter + adapt diag. I went through all the previous discussions on this topic and made sure the model was specified properly and that all my data was suitably preprocessed. It doesn’t blow up like this if I initialize …

You are right - it sometimes happen since the jitter might be too large if you have a RV on a small scale (currently it jitter uniformly around the test value). pymc-devs/pymc3/blob/0379ef57b2bcd65c6a2bee9eda13579b361019a4/pymc3/sampling.py#L1284-L1293 start = [] for _ in range(chains): …

Initialization energy is NaN or Inf with jitter

Questions

junpenglao February 27, 2018, 5:21am 4

We are currently trying to make the jitter more reliable https://github.com/pymc-devs/pymc3/issues/2856#issuecomment-367648655
Hopefully, this would get fixed soon!

Topic		Replies	Views
Difference between 'jitter+adapt_diag' and 'adapt_diag'? Questions	2	1271	April 2, 2019
Derivatives are zero for jitter but not ADVI initialization Questions	2	556	September 28, 2018
What exactly is `jitter+adapt_diag` and why is it the default now? Questions	3	3497	November 3, 2017
I get "Bad initial energy" even on one of the examples Questions	3	1161	March 6, 2019
ValueError: Bad initial energy: inf. The model might be misspecified error replicating Multilevel Regression and Poststratification with PyMC3 notebook Questions	2	1615	May 11, 2018

Initialization energy is NaN or Inf with jitter

Related topics